The processing cost of interpreting superlative modifiers and modals

Abstract

Superlative modifiers like 'at least' and 'at most' pose several challenges to formal semantic and pragmatic analyses. A particular challenge is accounting for the ignorance inferences they give rise to, and whether to attribute these inferences to the lexical semantics, the semantic combinatorics, pragmatic implicature, or the interaction thereof. We conducted a self paced reading study in order to adjudicate between the various analyses proposed for superlative modifiers, taking their interaction with deontic modals as a test case. In the experiment, we aimed to determine which superlative modifier-deontic modal combinations are correlated with which readings as well as the time-course of the interpretation of these expressions in order to better adjudicate between competing analyses. We found that some superlative-modifier combinations have a clear response pattern coupled with faster processing-time and argue that this is an indication that they give rise to an authoritative reading that comprehenders favoured. When such reading wasn’t available, participants resorted to a repair mechanism to generate an authoritative reading that is not compositionally available, which came at a processing cost.